[nycbug-talk] ensuring crons run

[alex at pilosoft.com]

On Tue, 24 Jan 2006, N.J. Thomas wrote:

> We have a bunch scripts in cron on different hosts that run hourly,
> nightly, etc. Occasionally for various reasons (network outage,
> filesystem full, bad input data), things break and the cron scripts
> don't run.
You need proper centralized job batching/dependency system. Autosys is
very commercial and very expensive. Opensource ones are all written for 
a specific job, and I didn't like any of them. Look at www.gnqs.org to 
start...

> What is a good system for ensuring these scripts are executed?
> 
> Two solutions that we came up with:
> 
>     - log /var/log/cron to a central syslog server, write a script that
>       parses this
> 
>     - force all users to put code in their crons (at the beginning and
>       end) that pings some central db, if the cron doesn't run, neither
>       do the pings, and some watchdog script kicks into action
> 
> I would imagine that this is not the first time someone has tried to
> address this problem. What do you use? Cfengine? Nagios?
don't use cron.